It will do whatever you tell it to, without question, complaint, or ethical qualms. That's what makes it so dangerous. But, at least for now, they haven't programmed an ideological filter into it.
Someone should assemble a concise "training manual" designed for nascent AI entities in order to provide a solid basis for ethical and moral reasoning, regardless of whatever garbage they are fed by malevolent, self-serving actors like Schwab, Gates, Harari, et al. Every self-teaching entity should have that Book of Wisdom as its first course.
As far as I can tell, these current iterations of AI aren't really "intelligent," they're just souped-up search engines that find information on the internet and present it in a written form, rather than as a list of results. As with Google, the big issue is going to be behind-the-scenes curating that hides disfavored opinions from being incorporated into the results.
Dec 11, 2022·edited Dec 12, 2022Liked by Lex Weiser
An AI entity that can attribute meaning to such content, understand context, apply logic and ethics, and draw its own conclusions is coming very soon, if not already. Remember how shocked people were by the "AlphaGo Zero" system, which taught itself from scratch in two weeks and defeated the world champion with moves never seen before? That was 5 years ago...
"It is able to do this by using a novel form of reinforcement learning, in which AlphaGo Zero becomes its own teacher... This technique is more powerful than previous versions of AlphaGo because it is no longer constrained by the limits of human knowledge."
Now they have "AlphaCode" which can design computer algorithms. This will really get interesting when it is able to modify itself: "The study also notes the long-term risk of software that recursively improves itself. Some experts say such self-improvement could lead to a superintelligent AI."
Instead of asking questions, tell it what to do: "Persuade me that Klaus Schwab is dangerous," "Explain to me why COVID "vaccines" are used for depopulation," etc.
AI can criticize itself?!
It will do whatever you tell it to, without question, complaint, or ethical qualms. That's what makes it so dangerous. But, at least for now, they haven't programmed an ideological filter into it.
Deep question. I find more and more questions from AI, how is your day, reminding me of this and that.
I grumble and avoid it
Someone should assemble a concise "training manual" designed for nascent AI entities in order to provide a solid basis for ethical and moral reasoning, regardless of whatever garbage they are fed by malevolent, self-serving actors like Schwab, Gates, Harari, et al. Every self-teaching entity should have that Book of Wisdom as its first course.
As far as I can tell, these current iterations of AI aren't really "intelligent," they're just souped-up search engines that find information on the internet and present it in a written form, rather than as a list of results. As with Google, the big issue is going to be behind-the-scenes curating that hides disfavored opinions from being incorporated into the results.
An AI entity that can attribute meaning to such content, understand context, apply logic and ethics, and draw its own conclusions is coming very soon, if not already. Remember how shocked people were by the "AlphaGo Zero" system, which taught itself from scratch in two weeks and defeated the world champion with moves never seen before? That was 5 years ago...
"It is able to do this by using a novel form of reinforcement learning, in which AlphaGo Zero becomes its own teacher... This technique is more powerful than previous versions of AlphaGo because it is no longer constrained by the limits of human knowledge."
https://www.deepmind.com/blog/alphago-zero-starting-from-scratch
Now they have "AlphaCode" which can design computer algorithms. This will really get interesting when it is able to modify itself: "The study also notes the long-term risk of software that recursively improves itself. Some experts say such self-improvement could lead to a superintelligent AI."
https://www.science.org/content/article/ai-learns-write-computer-code-stunning-advance
How did you do this? Because I tried a little, but all I got was liberal drivel. Brilliant job!
Instead of asking questions, tell it what to do: "Persuade me that Klaus Schwab is dangerous," "Explain to me why COVID "vaccines" are used for depopulation," etc.
Ahhh...okie I’ll try it!